UniWB affecting auto-exposure?

[hjulenissen]

Does UniWB affect auto-exposure? I.e. is it known if the AE algorithm of one or all current manufacturers is based on raw sensor level, "jpeg" level, or a combination?

-h

[stamper]

It affects the outcome. I used it for a while. As an experiment I took - of the same scene - some images and exposed to the right using the normal jpeg histogram. The information in the histogram was touching the right hand side. I then took the same images with the Uni WB. To expose using the Uni WB I had to add an extra 2/3 to 1 stop exposure to get the information in the histogram to touch the right hand side. The conclusion was the jpeg histogram was underexposing the information consistently by 2/3 to 1 stop. When I now use the jpeg histogram I know that there is a little more head room to expose. If I get the blinkies I can ignore them knowing I haven't over exposed.

[Bart_van_der_Wolf]

Does UniWB affect auto-exposure? I.e. is it known if the AE algorithm of one or all current manufacturers is based on raw sensor level, "jpeg" level, or a combination?-h

Who knows what all manufacturers do? However, AFAIK regular exposure determination takes place before an exposure will be made. I don't know of any automatic feedback for subsequent exposures (which might be of questionable use because lighting can change between exposures).

What will change the exposure is whether the regular exposure is manually adjusted with a +/- EV correction to expose-to-the-right (ETTR) without clipping. You can consider this as a fixed offset relative to the regular exposure metering.

Cheers,
Bart

[RFPhotography]

It affects the outcome. I used it for a while. As an experiment I took - of the same scene - some images and exposed to the right using the normal jpeg histogram. The information in the histogram was touching the right hand side. I then took the same images with the Uni WB. To expose using the Uni WB I had to add an extra 2/3 to 1 stop exposure to get the information in the histogram to touch the right hand side. The conclusion was the jpeg histogram was underexposing the information consistently by 2/3 to 1 stop. When I now use the jpeg histogram I know that there is a little more head room to expose. If I get the blinkies I can ignore them knowing I haven't over exposed.

Don't need UniWB for that.  Shooting RAW and seeing what head room one has in post serves the same purpose.  You can also adjust the JPEG preview provided by the camera on the LCD to better reflect that headroom. 

To the original question, WB is applied to the image after exposure with the in camera processor to provide the LCD JPEG preview.  Since WB is applied after exposure, it should have no impact on the metering of the scene.  What it's doing is altering the processed JPEG preview based on the exposure information, not altering the actual exposure.

[stamper]

I was under the impression that WB was baked into the RAW?

[digitaldog]

I was under the impression that WB was baked into the RAW?

WB has no role on the raw data, its just a metadata suggestion from the camera. Only ISO and exposure affect the raw data.

[bjanes]

I was under the impression that WB was baked into the RAW?

WB is merely added to the metadata attached to the file and the actual data in the raw channels is unaffected. If the camera allows for too much highlight headroom, you can also upload a custom tone curve to compensate for this.

Regards,

Bill

[stamper]

WB has no role on the raw data, its just a metadata suggestion from the camera. Only ISO and exposure affect the raw data.

But you have to make or accept some kind of adjustment? That was the thrust of my post. If you import a RAW image into lightroom or ACR it is there?

[RFPhotography]

Yes, there is a WB setting if you open a RAW image in ACR or LR.  But as has been pointed out it's based only on the metadata read from the image and isn't locked in at all.  That's the very essence and beauty of shooting RAW.  I wrote an article recently on my blog that discusses it, http://rf-photography.ca/power-raw/.  Nothing in depth but it gets to the issue.

This is different from a JPEG where the WB is cooked in.  And even with a JPEG captured in camera, the WB is 'added' after the exposure in processing.  The choice of WB setting has no impact on metered exposure.  

[stamper]

My point is that you can't process a raw without some kind WB setting whether it is in the metadata or not. My use of baked in may not have been the best description but the use of WB cannot be avoided?

Quote

To the original question, WB is applied to the image after exposure with the in camera processor to provide the LCD JPEG preview.  Since WB is applied after exposure, it should have no impact on the metering of the scene.  What it's doing is altering the processed JPEG preview based on the exposure information, not altering the actual exposure.

Unquote

You are technically correct but with respect to the final outcome when processing takes place in the ACR or Lightroom then the exposure is affected by you choice of WB setting?

[fdisilvestro]

What WB does is to multiply the numerical values of each channel by a factor. Usually the green channel is left at "1" and the blue and red channel are multiplied by a value different than "1".

When you adjust exposure in post, you are basically multiplying all channels by the same factor.

It might be possible that a particular WB cause a channel to clip and you have to adjust exposure (or recovery in LR)

The important issue is to have the original RAW data unclipped. Clipped RAW values mean lost data.

[Guillermo Luijk]

you can't process a raw without some kind WB setting whether it is in the metadata or not. My use of baked in may not have been the best description but the use of WB cannot be avoided?

You can develop a RAW file without applying a WB on it, just need to use a RAW developer that allows this. However, why should you want an image with no WB applied? they look greenish because of the G channel usually being the one getting the higher values in the RAW data.

Left image was obtained without applying WB. Right image was obtained with destructive Daylight WB (overexposing WB) preset:



The same RAW file with a preserving Daylight WB (no overexposing of RAW data), allows to get the best possible image:




Regards

[ErikKaffehr]

Hi,

Image is usually processed in 16 (or 15) bit space while raw data contains mostly 14 bits, this gives essentially a headroom of 1-2 stops to manipulate data.

BR
Erik

Yes, there is a WB setting if you open a RAW image in ACR or LR.  But as has been pointed out it's based only on the metadata read from the image and isn't locked in at all.  That's the very essence and beauty of shooting RAW.  I wrote an article recently on my blog that discusses it, http://rf-photography.ca/power-raw/.  Nothing in depth but it gets to the issue.

This is different from a JPEG where the WB is cooked in.  And even with a JPEG captured in camera, the WB is 'added' after the exposure in processing.  The choice of WB setting has no impact on metered exposure.